PANS prediction of passenger vehicle flows

Abstract

This thesis was performed at Chalmers University of Technology in cooperation with AVL. The Partially Averaged Navier-Stokes model that was investigated here has the opportunity of being a very powerful tool with promises of being accurate and relatively low cost. In this thesis the PANS model is evaluated for the flow over an SAE 30 degree notchback reference body which is a simplified representation of a car. Two different discretization schemes have been used here and compared to data from tests done on the same model in a wind tunnel. The results show a fairly good agreement with respect to the flow structures and the SMART simulation shows very good agreement with respect to drag and base pressure. The SMART simulation shows many more resolved small flow structures than the MINMOD simulation but also shows separation at a too early point. These results can be summarized as that the model has good potential and is good at predicting flow structures and drag; there is however still a significant amount of work to be done with choosing discretization and investigating the mesh resolution impact on a case like this. Further work on more complex shapes also has to be done however that was not possible in this thesis due to the very difficult meshing in AVL Fire.